Programming codes are traditionally based on binary numbering systems–everything is expressed in values of 0 or 1 in line with the true or false principles of classical logic. Fuzzy logic is different in that it takes a non-binary approach and allows for infinite values between 0 and 1.
Andreas Meier, emeritus professor of data science at the Faculty of Economics and Social Sciences of the University of Fribourg in Switzerland, is on a mission to raise awareness of the importance of fuzzy-based computing for business, government and society.
Read more about his research in Fuzzy Management Methods, here: https://doi.org/10.1007/978-3-030-03368-2
Read more about the research at: FMsquare Foundation
Image source: graphicwithart / shutterstock.com
Transcript:
Hello and welcome to Research Pod. Thank you for listening and joining us today.
In this episode we look at the work of Andreas Meier, emeritus professor of data science at the Faculty of Economics and Social Sciences of the University of Fribourg in Switzerland. Andreas is an expert in fuzzy logic and he’s on a mission to raise awareness of the importance of fuzzy-based computing for business, government and society.
Before we explore what that means, Andreas ranks fuzzy logic as one of the seven wonders of the IT world. The others include the computer mouse for pointing to and controlling objects on screen, databases which allow us to store and retrieve information, and cryptography to protect users’ privacy. Also on the list are the graph grammars used by computer programmers to perform concepts and relationships, internet browsers which allow us to search hypertext worldwide, and blockchain, which is in essence a database for distributed and decentralised accounting and asset management.
The wonder of fuzzy logic, according to Andreas, is that it introduces intuition into the world of IT and brings human and artificial intelligence together. That’s not to say that logic is fuzzy, but rather to acknowledge that data can be fuzzy, and IT can work with it in new ways, to everyone’s benefit.
To understand more, we need to think about how computers work. The programming codes traditionally used to perform instructions on digital computers are based on binary numbering systems – everything is expressed in values of 0 or 1 in line with the true or false principles of classical logic. Fuzzy logic is different in that it takes a non-binary approach and allows for infinite values between 0 and 1. Shades of grey begin to enter a world that has previously been seen in black and white.
The beauty is that, because data no longer has to be reduced to binary absolutes, IT can permit grades of truth. In this way IT can better resemble the reality of human behaviour and experience.
This is especially important in our increasingly complex world. In the age of big data, it’s becoming more and more difficult to analyse all the information available. Management decisions are informed, for example, by everything from emails, social media and internet searches, to content management and geographical information systems, as well as customer databases, stock exchange data, and data from the electronic measurement of anything from energy use to manufacturing inputs.
The problem isn’t only information overload. It’s also how to assess the quality and significance of all the data available. Data can sometimes be misleading, undefined, inaccurate or uncertain. It also includes qualitative as well as quantitative elements. As Andreas argues, truth is graded – everything is true to a certain degree. The solution therefore lies in developing procedures that evaluate and grade the truth of the data available to us.
This is where fuzzy logic comes in. The concept was proposed by Lotfi Zadeh of the University of California at Berkeley in the 1960s and is based on set theory – the branch of mathematics that describes collections of things with the same characteristics as ‘sets’.
Classical set theory sees objects in binary terms, according to whether their membership of a given set is true or false. In contrast, fuzzy set theory holds that membership of a set can be graded according to a scale of infinite degrees of truth.
Andreas argues that by thinking of graded truth, rather than binary opposites, we can use IT to improve the quality of decision-making, and this has significant benefits for business, government and society.
For example, consider a business that wants to assess the value of its individual customers.
The goal of customer relationship management is for businesses to personalise their response to individual customer requests and behaviour. This is on the understanding that customers are assets, and should be treated according to their market and resource potential. This in contrast to the traditional approach of segmenting customers into sharply defined classes, and responding to them based only on that information. The problem is that this treats all customers within a class equally.
For example, consider a customer portfolio based on only two assessment criteria: turnover in Euros, with value ranges of low or high; and loyalty, with value ranges of bad, good, weak or great. If classes are sharply defined, it may be hard to spot customers with development potential, because every customer in the same class receives the same rating. On the other hand, if classes are not sufficiently well defined, customers may not all be evaluated correctly.
Andreas argues that such risks can be mitigated if fuzzy classes are used to expand the value ranges in the business’s initial assessment of its customers’ value. For example, the business might grade customers with various specific revenues, and with loyalty attributes on a range from bad, through weak and good to great.
Customer value could then be calculated more precisely by aggregating qualitative and quantitative affiliation values and help to identify which customers have the best development potential, as well as those the business is most at risk of losing. In addition, as each customer has an individual assessment value, the company can target its marketing spend and manage its customer relationships individually.
Andreas argues that politics is another area in which fuzzy logic can help.
Political crises often develop when parties gain power, or decisions are made, according to a narrow margin of votes. The result forces a division into winners and losers that can lead to polarised societies and cause voters to lose faith in democracy.
Take the example of the UK’s 2016 Brexit referendum in which 51.9% of voters voted to leave the European Union and 48.1% voted to remain. The result was of great significance to the UK and to Europe, yet the vote was carried by a margin of just 3.8%. The outcome also failed to take account of significant differences between voting patterns in England and Wales, the majority of whom wanted to withdraw from the EU, and Northern Ireland and Scotland, the majority of whom wanted to stay.
Andreas argues that electoral systems that are based on binary yes or no choices ignore the complexity of human decision-making. A fuzzy logic approach that takes account, for example, of those who are half-for and half-against a proposal (50 – 50), as well as those who are 25% or 75% in favour, would achieve more balanced outcomes.
Differentiated thinking, that is applying fuzzy logic, can also help with the difficult topic of how to ensure that products and services based on artificial intelligence are ethically sound.
Andreas argues that, while such things as drones and other computer-generated devices and systems are not in themselves intelligent or ethical, it is possible to evaluate their effects according to moral and ethical principles. A fuzzy approach can help because it allows humans to interact with computers using words instead of numbers. Not only is this closer to human experience, it is particularly suitable in situations where meaning is imprecise.
Andreas, working with one of his former PhD students, proposes a ‘radar’ system of ethics evaluation based on human rights as recognised by the United Nations. The six axes of their radar comprise equality and justice, protection and assistance, privacy and data protection, as well as freedom and autonomy, solidarity and sustainability, and accountability and transparency.
By modelling these six dimensions using linguistic variables, Andreas argues that it’s possible to aggregate values and produce formulas to calculate and grade the ethics of software systems. Not only does this apply to the example of drones, it can also help with the evaluation of other products and systems, for example robot applications, decision algorithms, big data analytics, business processes and web services.
The three examples from business, politics and ethics show how soft computing – that is developing human-oriented algorithms based on fuzzy logic – can be used by companies, government and society to build trust in today’s digital economy. Fuzzy-based reasoning allows for human-centred digital transformation and can help to produce digital tools that complement human intelligence, rather than replace it.
Andreas argues that the challenge is now to tackle what his research colleagues Radim Belohlavek from the Czech Republic, and Joseph Dauben and George Klir from the USA, have called the “grand paradigm shift”. What is needed is a change in thinking from the binary values that have dominated logic and mathematics for millennia to the multi-values that more accurately represent human experience in the 21st century. Andreas believes that graded truth has the power to help leverage collective and individual intelligence for a human-centered digital transformation. It’s time for change and fuzzy logic’s time has come.
That’s all for this episode – thanks for listening, and stay subscribed to Research Pod for more of the latest science and ideas. See you again soon.
Leave a Reply